Assessing symptoms before hysterectomy: is the medical record accurate?

Abstract

Our purpose was to evaluate the agreement between the documentation of symptoms leading to hysterectomy and the assessment of those symptoms by the patient. A retrospective study was performed of 497 women in southern California who had hysterectomies. Sensitivity, specificity, and kappa statistics were calculated for the medical records and were compared with patient interviews for the presence and severity of symptoms. The medical record was 93% sensitive and 61% specific for identifying bleeding and 79% sensitive and 55% specific for identifying pain. Overall agreement between physician records and patient interviews was moderate for bleeding (kappa, 0.55-0.58), fair for pain (kappa, 0.29-0.34), and poor for impairment as a result of bleeding or pain (kappa, 0.0-0.14). Physician overestimation of symptoms could lead to overuse of hysterectomy, whereas underestimation could result in underuse. Our results suggest that both underestimation and overestimation occur for patients with abnormal bleeding, pain, or both. If physicians accurately assess symptoms but fail to document them, examinations of appropriateness will be faulty unless patients are interviewed.